New York didn’t brainstorm this move, it banged it on the table. The state is signaling that the era of “move fast and hope nothing breaks” in AI may finally be wearing thin, with Gov.
Kathy Hochul signing the Responsible AI Safety and Education Act – or RAISE Act, for short. The law homes in on powerful, cutting edge AI systems and demands something many critics have said has been missing for years: actual accountability, as opposed to glossy ethics statements.
The legal breakdown that first detailed just how far-reaching this revolution could be was discovered in analyzing the legislation’s very structure and, by extension, New York State’s very intention to move beyond mere transparency and into a posture of actual oversight.
At the core of the RAISE Act stands an uncomfortable question for AI developers: What do you do when things start going wrong?
Companies that develop or use high-risk AI systems will also need to keep safety documentation, provide regular transparency reporting and alert authorities if their AI system does any damage.
That timeline alone has raised eyebrows in Silicon Valley, where disclosures typically take weeks, not days.
Legal analysts say the regulation marks a sea change in how AI risk is framed – as something less akin to a public relations problem and more like a public safety concern.
To make the move even more intriguing, New York isn’t acting alone. California has already released its own frontier AI law, and comparisons are likely.
Although both states are looking to clamp down on runaway A.I. threat, their approaches vary in how sweeping they are, in how they would be enforced and in the time frame for those regulations – nitty-gritty details that can pack a big wallop if you’re a corporation operating across this vast land of ours.
Some legal observers contend that businesses could be forced to design systems that meet the toughest state rules, if only for their own mental health.
A head-to-head legal comparison illuminates both how New York’s RAISE Act measures up to California’s and where the real pressure points are.
Zoom out, and the timing gets messier still. The RAISE Act comes as Washington is still grappling with the question of whether AI should be regulated at a national level, a local one – or somewhere in between.
Already, a recent federal call for more AI regulation has triggered discussions about whether state-level rules could clash with future national standards.
Compelling as it is to look away from the tension, companies are not looking away. Legal experts caution that the second era of A.I. oversight could become a jurisdictional tug-of-war, with states like New York taking the lead while federal agencies race to follow.
Here’s my own view, for what it’s worth: The RAISE Act doesn’t strike so much like an anti-innovation hammer as a skimpy reality check.
AI is no longer simply driving chatbots – it’s affecting decisions in healthcare, financial markets and public infrastructure.
It is naïve at this point to pretend that voluntary guardrails are sufficient. Certainly, there will be pain in complying, and yes, a few start-ups are going to make the most noise.
But it’s not exactly unreasonable when the stakes are this high to ask developers to justify that they’ve thought through worst-case scenarios.
New York has issued a challenge: Develop powerful AI - but do it responsibly, and be prepared to explain how you did it.
What the American response evolves into may serve as a national blueprint or simply a cautionary tale. Either way, the AI safety discussion is now starting for real, and there’s no turning it back.